1. Introduction

The development of integrated circuits in the semiconductor field is closely associated with the technology advances of nanomanufacturing. Photolithography, the most commercialized technology, has not only formed the cornerstone of IC industrial, but also been widely employed in micro/nano optics [1], photonic crystals [2] and metasurface devices [3], etc. However, hindered by the optical diffraction limitation, the resolution is subject to the feature size D = k₁λ/NA, where k₁ is the process factors, λ is the lithography wavelength and NA is the numerical aperture. Compared with off-axis illumination [4], proximity-effect correction [5] or immersion lithography, shrinking the wavelength of illumination light is the most straightforward method to achieve higher resolution. Photolithography based on light sources such as 193 nm from ArF excimer lasers, 13.5 nm from extreme ultraviolet radiation or soft x-ray [6] have been significantly expended over the decades. The shorter wavelength light source, however, will result in complex system and high cost, which is unfavorable for nano-fabrication. The alternative lithography techniques, e.g., electron beam lithography [7] and focused ion beam [8] are facing the similar problem, and it is hard to balance the accuracy and the cost. Therefore, a high-accuracy and low-cost nanomanufacturing technology is still highly desired.

Surface plasmons (SPs), which are surface waves excited at the interface of metal and dielectric, were recently employed for investigating plasmonic lithography, e.g., interference lithography [9,10]. By converting the space light into propagating SPs and inducing the localized SPs, it is able to break the fundamental limit of traditional optical diffraction and confine the light at deep-subwavelength scale [11–15]. Nanoscale patterning with half-pitch resolutions less than 50 nm have been confirmed using 365 nm wavelength illumination [16,17], demonstrating that plasmonic lithography is a viable and low-cost alternative lithography technology with high resolution. According to the operating mechanism, there are two major modalities of plasmonic lithography, the projection lithography [18,19] and the direct-write lithography [20,21]. The former has relatively higher reproducibility and allows for large-area manufacturing, but relies on a 1:1 mask. While the latter is able to fabricate arbitrary structures via mechanically scanning without requiring the mask and provides a more flexible method. However, both categories suffer from the challenge of shallow exposure depth due to the exponential decay of SPs in the perpendicular direction. PSLs, such as plasmonic superlens [22], reflective plasmonic lens [23,24], and plasmonic resonant cavities [25,26] are proposed to amplify and compensate evanescent waves, and employed in projection lithography to improve depth of field. Such strategies could also be used in the direct-write scheme by incorporating the PSLs, but the relevant investigations are rarely reported.

In this paper, we design and fabricate two plasmonic probes with different localized SP resonant mechanisms. Two common ultraviolet lasers with wavelengths of 405 nm and 355 nm are considered as the light sources shining onto the probing tips. The PSLs are designed and optimized from the aspect of materials and structures to increase the field confinement and the depth of focus, which are critical to PDWL. We emphasize that different PSLs and illumination wavelengths are required for different plasmonic probes. The best configuration of PSL and illuminating wavelength for one probe is not applicable to another and the improper configuration may even accelerate the light decay and divergence into the photoresist. For the probe with bowknot aperture, the combination of the Ag-Ag cavity PSL and 355 nm illumination presents the best enhancement of depth of focus, while for the probe with asymmetric ring slit, it would be preferable to employ the Ag-Al cavity PSL and 405 nm illumination. The proper PSLs can effectively amplify and compensate evanescent waves, resultantly improving the contrast, exposure depth of resist patterns and tolerance to the air gap. This work would help to improve the resolution and exposure depth of PDWL.

2. Results and discussion

2.1 Scanning probe lithography with plasmonic probe and plasmonic slab lens

Two plasmonic probes working under the illumination of linear polarized beam are employed to excite local SPs and focus the light into deep subwavelength spots at the probing tips as shown in Fig. 1. The probe is modeled from the commercial AFM probe (qp-fast, Nanosensors) made of a quartz-like material. For the probe with bowknot aperture, the tip is truncated with a flat surface in 250 nm diameter on top, a thin Cr layer of 60 nm thicknesses is deposited on the tip, and finally a bowtie nano-aperture is designed on the top metal film [Fig. 1(a)]. The incident light impinges the probe from the cantilever side and is guided to the tip, inducing localized surface plasmon (LSP) in the bowknot aperture. The electric is concentrated inside the gap of the bowknot, which is dominated by the in-plane electric field [27] and can be directly excited by the linear polarization component of illumination. While for the probe with asymmetric ring slit, a thin Cr layer of 60 nm thicknesses is deposited on the original probing tip and a semi-circular ring is designed on the Cr film [Fig. 1(b)]. This is different from the case of the bowknot aperture. The SP waves are first excited by the semi-circular ring slit and then guided to the tip apex to excite the LSP, where the hotspot is dominated by the out-of-plane electric field [28]. Figure 1(c) and 1(d) plot the details of the nano features of the above two probes with the optimized dimensional parameters: a = 120 nm, b = 160 nm, c = 30 nm, g = 30 nm, and w = 70 nm. The scanning electron microscopy images of these two fabricated probes are shown in Fig. 1(e) and 1(f), respectively. The flat surface at the tip apex is milled using Zeiss triple-beam focused-ion-beam (FIB, zeiss orion nanofab, Jena, Germany) microscope. 60-nm-thick Cr film is deposited on the tip by electron-beam evaporation (anelva L-400EK, Kawasaki, Japan). Considering the geometry difference of the bowknot and semi-ring, the former is milled with He-FIB to process the sharp ridge, while the latter is fabricated with a high-energy Ga-FIB.

 

Fig. 1. The direct-write lithography with plasmonic probe and plasmonic slab lens. Designs of the probing tips with a (a) bowknot aperture and (b) an asymmetric ring slit. The geometries of the (c) the bowknot aperture and (d) the asymmetric ring slit. (e), (f) The scanning electron microscope images of these two fabricated probing tips. Schematic diagrams of the plasmonic lithography (g) without PSL and (h) with PSL.

Download Full Size | PDF

Generally, PDWL is capable of fabricating arbitrary patterns via mechanically scanning a probe without relying on the masks. The high intensity LSP confined at the tip apex is utilized to expose the photoresist in the near-field condition [Fig. 1(g)]. Though the probing tip is located tens of nanometers above, or even in contact with the photoresist, however, the LSP is subjected to dramatic decay and strong divergence once emerging from the nano-aperture or the nanostructure. To improve the exposure depth of PDWL, we incorporate the plasmonic layers into the photoresist substrate [Fig. 1(h)] and investigate three kinds of PSLs, the plasmonic lens (PL, a metal film above the photoresist), the plasmonic reflector (PR, a metal film underneath the photoresist) and the plasmonic cavity (PC, two metal film separated by the photoresist layer). Two low-loss metals in ultraviolet range, Ag and Al are considered as the materials of metal films. The permittivities of Cr, Ag, Al and silica are referred from Palik’s handbook [29]. The permittivity of the commercial photoresist (AR-P 3170, Allresist, Germany) ε_pr= 2.7 is obtained by the spectroscopic ellipsometry (V-VASE, J.A.Woollam, USA). The thicknesses of the PL, photoresist, and PR are 20 nm, 30 nm and 70 nm, respectively (see Section I of Supplement 1 for more information about the optimization of the film thickness). We note that since the introduction of PSLs, further pattern transfer after exposure and development can be performed via the hard-mask technology and multilayer etching process, which have been demonstrated to increase the aspect profile depth of the lithography patterns [25], while the residual metal film can be removed by ion beam etching.

2.2 Scheme optimization for probe with bowknot aperture

We first investigate PDWL for the probe with bowknot aperture and optimize the exposure depth into photoresist via PSLs. To visualize the excitation of LSP and its divergence into the photoresist, 3D full-wave simulations in the time domain module of the commercial software Computer Simulation Technology (CST) is performed by considering a x-polarized beam normally impinging on the probing tip. The mesh size of simulation is set to be 2 nm × 2 nm around the tip apex area. A 355-nm-wavelength beam is normally impinging the tip from the top and Fig. 2 shows the comparison of electric field distributions with and without a PSL. An air gap of 10 nm is set between the tip and the substrate. The lithography configurations without PSL, with a PC and with a PR are displayed in Fig. 2(a), (d) and (g), respectively (see Section II of Supplement 1 for plasmonic lithography with a PL). Here Ag is the only material selected for both the PC and the PR, and the material optimization will be illustrated later. As expected, electromagnetic energy is concentrated inside the gap of bowknot aperture, inducing significant local field enhancement. However, the light diverges very fast into the photoresist without PSL as shown in Fig. 2(b). Figure 2 (c) plots the electric field profiles along the x direction at top, middle and bottom positions of the photoresist layer. The intensity at bottom layer is much lower than that at top, indicating the low penetration. In contrast, when the photoresist layer is filled into a PC, or a PR is added below the photoresist layer, the light is effectively coupled into photoresist and generates quite confined hotspot as shown in Fig. 2(e), (f) and (h), (i). These phenomena can be attributed to the compensating and amplifying of evanescent waves by the PR. Besides, the PC further enlarge the in-plane component E_x and reduce the out-of-plane component E_z [25], while the latter component is unfavorable for the bowknot aperture tip and would extend the focus spot.

 

Fig. 2. Lithography simulation for the probe with bowknot aperture illuminated by a 355-nm-wavelength x-polarized beam. (a) The schematic of PDWL without PSL. (b) Electric field amplitude distributions in the cross section parallel to the incident polarization. (c) The one-dimensional electric field amplitude distribution along x-axis. (d)–(f) The schematic of PDWL with a PC and the electric field amplitude distributions. (g)–(i) The schematic of PDWL with a PR and the electric field amplitude distributions.

Download Full Size | PDF

We use the imaging contrast function C(z, r) = [I_max(z)-I(z, x)]/ [I_max(z)+ I(z, x)] to quantify the lithography resolution, where z is the distance from top layer of the photoresist, I_max(z) is the peak intensity of the light at distance z and I(z, x) is the intensity at position (z, x) [26]. The contrast distributions for the three cases in Fig. 2 can then be calculated with the simulation data. Considering an imaging contrast >0.2 is enough to produce a satisfactory exposure, we can obtain the achievable resolution at different depths in the photoresist by calculating the x positions where C(z, r) is equal to 0.2. The resolutions at top, middle and bottom positions are 52 nm, 62 nm and 89 nm for the conventional lithography in Fig. 2(a), 21 nm, 29 nm and 27 nm for the PC lithography in Fig. 2(d), and 40 nm, 32 nm and 30 nm for the PR lithography in Fig. 2(g), respectively. At the bottom of the photoresist, the achievable resolution of the lithography with PSLs is one-third of that without the PSLs, which indicates low contrast and ultra small pattern depth in conventional PDWL can be greatly improved benefitting from the PSLs. We note that the resolution can be further improved by reducing the air-gap distance or applying the contact mode.

The materials and illuminating wavelength are further optimized for the bowknot aperture tip. Two common ultraviolet lasers with wavelengths of 405 nm and 355 nm are chosen as the light sources shining onto the probing tips, and two low-loss metals, Ag and Al are considered as the materials of PSLs. Figure 3 shows the simulation results for the PR and the PC with different metal materials and different illuminating wavelengths, where one-dimensional electric field amplitude distributions along both x-axis and y-axis at top position of the photoresist layer are plotted to estimate the whole spot size. It is shown that both the peak intensity and the spot size are affected by the material selection. For the PR configuration in Fig. 3(a) and 3(b), the Ag reflector exhibits higher peak intensity and smaller spot size than Al reflector, and the 355-nm-wavelength beam is more effectively focused than the 405-nm-wavelength beam, which can be ascribed to the matching of wave vector. The permittivity of Ag at 355 nm (ε_Ag= -2.03 + 0.60i) approaches to that of photoresist, thus effectively restraining the LSP from diffusion. While the Al reflector contributes less to the local field enhancement and confinement due to much larger permittivity of Al than that of photoresist. Such difference between Ag and Al PSLs is more significant in the PC structure as shown in Fig. 3(c) and 3(d). The Ag-Ag PC (photoresist layer sandwiched by a top Ag film and a bottom Ag film) exhibits much better performance than other PSL structures especially at 355 nm wavelength and compresses the light into a smaller spot by enhancing the main component E_x with high light intensity and damping E_z component that makes negative contribution [12].

 

Fig. 3. Optimization of PSLs for the bowknot aperture tip. The one-dimensional electric field amplitude distributions along x-axis and y-axis at top position of the photoresist layer are compared for PSLs with different metal materials and different illuminating wavelengths. (a) Ag and Al PR, 355 nm wavelength. (b) Ag and Al PR, 405 nm wavelength. (c) Ag-Ag, Ag-Al, Al-Ag and Al-Al PC, 355 nm wavelength. (d) Ag-Ag, Ag-Al, Al-Ag and Al-Al PC, 405 nm wavelength.

Download Full Size | PDF

For further quantitative analysis, Fig. 4 gives the full width of half maximum (FWHM) of the spot size and the peak intensity along z direction for different PSL configurations and illuminating wavelengths. In the conventional lithography without PSL [Fig. 4(a) and 4(b)], the light decays and diverges very fast into the photoresist, while the results of lithography with PSLs depend on the operating wavelengths. At 355 nm wavelength, the PR confines the light and compress the spot size along x-direction, and the spot size along y-direction remains almost unaffected [Fig. 4(c)]. Although amplified by the PR, however, the light intensity is greatly attenuated from top to bottom layer of the photoresist. The PC effectively improve this issue, where the attenuation is slowed down and the spot size is also nearly invariant in photoresist along the depth direction [Fig. 4(e)]. Differently, the PSL hardly works at 405 nm wavelength. The spot size is only slightly reduced and the light intensity decays rapidly for the PR [Fig. 4(d)], while the spot is even enlarged along x-direction for the PC [Fig. 4(f)].

 

Fig. 4. Peak intensity and FWHM as the functions of the depth into the photoresist for the bowknot aperture probe with different substrate configurations and illuminating wavelengths. The gap between the tip apex and the substrate is fixed as 10 nm. (a) without PSL at 355 nm. (b) without PSL at 405 nm. (c) Ag PR at 355 nm. (d) Ag PR at 405 nm. (e) Ag-Ag PC at 355 nm. (f) Ag-Ag PC at 405 nm.

Download Full Size | PDF

By adding the PSL, not only the resolution and depth of PDWL, but also the tolerance of air gap can be improved significantly in comparison with the structure without PSL. Figure 5 further provides the simulated results to evaluate the influences of separation air thickness on the spot size. The calculated results show that the spot size gradually increases with the increase of air gap, and the intensity declines to less than 1/10 when air gap increases from 10 nm to 50 nm [Fig. 5(a) and 5(b)]. For the lithography with PSLs, the spot has higher intensity and smaller size for the same air thickness. At 355 nm wavelengths illumination, the intensity declines to around 1/5 for the 50 nm air gap and the decay rate is reduced by 50% [Fig. 5(c) and 5(e)]. In addition, the spot size for PC lithography moderately varies with the air gap and shows best tolerance of air gap. At 405 nm wavelength illumination, the intensity decay rate with air gap shows similar feature [Fig. 5(d) and 5(f)], but the field intensity is not obviously improved. The above simulations testify to the remarkable advantage of PSL in PDWL. The results suggest that Ag-Ag PC and 355 nm illumination is the best combination for the lithography with bowknot aperture probe, which could effectively guarantee exposure depth and lithography resolution.

 

Fig. 5. Peak intensity and FWHM as the functions of the gap between the tip apex and the substrate for the bowknot aperture probe with different substrate configurations and illuminating wavelengths. The depth position into the photoresist is fixed as 5 nm. (a) Without PSL at 355 nm. (b) Without PSL at 405 nm. (c) Ag PR at 355 nm. (d) Ag PR at 405 nm. (e) Ag-Ag PC at 355 nm. (f) Ag-Ag PC at 405 nm.

Download Full Size | PDF

2.3 Scheme optimization for probe with asymmetric ring slit

We then simulate the PDWL for the probe with asymmetric ring slit and perform the similar optimization of PSLs to improve the exposure depth. For simplified calculation and comparative analysis of the focusing performance, Fig. 6 gives the simulated light field distributions of probing tip for different lithography configurations. A x-polarized beam with wavelength of 355 nm is normally impinging the tip from the top, and the air gap between the tip and the substrate is still fixed as 10 nm. The inner radius of the semi-ring is optimized as r = 450 nm. Here we compare the electric distributions with three kinds of PSLs, the PL [Fig. 6(a), 6(b) and 6(c)], the PR [Fig. 6(d), 6(e) and 6(f)] and the PC [Fig. 6(g), 6(h) and 6(i)], where Ag is selected as material for the PL and the top metal film of PC, Al is selected as material for the PR and the bottom metal film of PC. The semi-ring excites SPs and boost the energy accumulation at the tip apex. It is observed that the evanescent waves are amplified by the PL, delivering to the local field enhancement. However, the light intensity decay dramatically when penetrating into the photoresist and exhibits shallow exposure depth. In contrast, the PR compensates the loss of evanescent waves by totally reflecting the light passing through the photoresist, preserving the peak intensity almost non-decaying. However, the high sidelobes in the vicinity of the focal spot decrease the contrast and lead to poor field confinement. The PC, which is the combination of a PL and a PR, successfully solves the problems and succeeds the advantages of these two PSLs. The peak intensity in photoresist along the depth direction is nearly invariant, which could effectively guarantee exposure depth for the PDWL. The simulation results indicate that the PL confine and refocus the local electric fields, the PR keeps the field uniformity in the longitudinal direction, while their combination could greatly relieve issues of low contrast and ultra small pattern depth in conventional plasmonic lithography. We also calculate the lithography resolution of the semi-ring probe. The resolutions at top, middle and bottom positions of photoresist layer are 16 nm, 20 nm, and 23 nm, respectively, for the lithography configuration with a PC. It should be noticed that the near-field enhancement and resolution strongly depend on the radius of the tip apex. A conservative radius of 60 nm is used in our simulation. In fact, a much sharper tip with radius less than 10 nm has been experimentally obtained [13], which would further compress the focusing spot and increase the local field at tip apex.

 

Fig. 6. Lithography simulation for the probe with asymmetric ring slit illuminated by a 355-nm-wavelength x-polarized beam. (a) The schematic of PDWL with a PL. (b) Electric field amplitude distributions in the cross section parallel to the incident polarization. (c) The one-dimensional electric field amplitude distribution along x-axis. (d)–(f) The schematic of PDWL with a PR and the electric field amplitude distributions. (g)–(i) The schematic of PDWL with a PC and the electric field amplitude distributions.

Download Full Size | PDF

To evaluate the influence of the PSL material and illuminating wavelength on the spot size, the one-dimensional electric field distribution of different substrate structures at 355 nm and 405 nm wavelengths are simulated and plotted in Fig. 7. It should be noticed that the optimal inner radius r is different for different illuminating wavelengths, which is determined by comprehensive consideration of the slit length, the SP propagating loss, and the interference via multi reflection between the tip apex and the slit. The optimal radius is r = 450 nm for 355 nm wavelength and r = 350 nm for 405 nm wavelength. Different from the bowknot aperture probe, the focused spot is mainly ascribed to longitudinal field component E_z, while in-plane component enlarges the spot size and makes negative contribution. Therefore, the ideal PSL should only amplify the longitudinal component E_z. We compare the field distributions of different components for the photoresist substrate without PSL [Fig. 7(a) and 7(b)], with a PL [Fig. 7(c)–7(f)] and with a PR [Fig. 7(g)–7(j)]. It can be concluded that, for the PL, the Ag film amplifies both the in-plane and out-of-plane components, while in opposite, the Al film damps both field components. Due to large permittivity, the Al film reflects most of the incident light and is therefore not suitable for the PL. For the PR, the Ag film amplifies the in-plane component E_x and damps the out-of-plane component E_z, which would improve the bowknot aperture probe, but would extend the spot and exhibit the worst performance for the semi-ring probe. While the Al PR amplifies the in-plane component but has little impact on the out-of-plane component. The results confirm that both the peak intensity and the spot size are greatly affected by the material selection of PSL. It would further provide guide for the design of PC, that is, Ag for the top metal film and Al for the bottom metal film [Fig. 6(g)–6(i)].

 

Fig. 7. Optimization of PSLs for the semi-ring probe. The electric field components E_x, E_y and E_z distributions along x-axis at top position of the photoresist layer are compared for PSLs with different metal materials and different illuminating wavelengths. (a) without PSL, 355 nm wavelength. (b) without PSL, 405 nm wavelength. (c) Ag PL, 355 nm wavelength. (d) Ag PL, 405 nm wavelength. (e) Al PL, 355 nm wavelength. (f) Al PL, 405 nm wavelength. (g) Ag PR, 355 nm wavelength. (h) Ag PR, 405 nm wavelength. (i) Al PR, 355 nm wavelength. (j) Al PR, 405 nm wavelength.

Download Full Size | PDF

Figure 8 gives the spot size of FWHM and the peak intensity along z direction for further quantitative analysis. Different PSL configurations and illuminating wavelengths are compared to find out the best lithography performance for the semi-ring probe. The PR lithography is not considered here for its poor contrast. In conventional lithography without PSL [Fig. 8(a) and 8(b)], the light decays and diverges very fast into the photoresist, while the results of lithography with PSLs depend on the operating wavelengths. At 355 nm wavelength, the PL confines the light and reduces the spot size along both x- and y-directions [Fig. 8(c)]. However, the light intensity is still greatly attenuated from top to bottom layer of the photoresist. Although the PC effectively improve this issue, where the attenuation is slowed down and the spot size is nearly invariant in photoresist along the depth direction [Fig. 8(e)], but the intensity is relatively weak. At 405 nm wavelengths illumination, the intensity decay and spot size variation with depth show similar features with that at 355 nm wavelength except that the field intensity is greatly enhanced [Fig. 8(d) and 8(f)]. Specially, the light intensity at top and bottom layer of photoresist is enhanced by a factor of 2.5 and 4 for PC lithography compared to that without PSL. The results suggest that Ag-Al PC and 405 nm illumination is the best combination for the lithography with semi-ring probe. Although the optimized PSL structure and operating wavelength are different from that for bowknot aperture probe, this method also exhibits remarkable advantage to guarantee exposure depth and lithography resolution. Moreover, we emphasize that the peak intensity for the semi-ring probe can be further increased by designing more asymmetric semi-circular rings on the Cr film or replacing Cr with other metals such as Al, while the spot size and intensity decay in the photoresist remain almost unchanged. Here we choose metal Cr because of its good mechanical properties, which can protect the tip from damage when scanning tens of nanometers above the substrate.

 

Fig. 8. Peak intensity and FWHM as the functions of the depth into the photoresist for the semi-ring probe with different substrate configurations and illuminating wavelengths. The gap between the tip apex and the substrate is fixed as 10 nm. (a) Without PSL at 355 nm. (b) Without PSL at 405 nm. (c) Ag PL at 355 nm. (d) Ag PL at 405 nm. (e) Ag-Al PC at 355 nm. (f) Ag-Al PC at 405 nm.

Download Full Size | PDF

3. Conclusion

In summary, we design and fabricate two plasmonic probes with different localized SP resonant mechanisms, and investigate their lithography performance in combination with different PSLs. The peak intensity, spot size and decay rate of local light field at the tip apex are optimized by considering the illuminating wavelength, material and structure of the PSLs. Our simulation results suggest that Ag-Ag PC and 355 nm illumination is the best combination for the lithography with bowknot aperture probe, while the semi-ring probe exhibits better performance under the condition of Ag-Al PC and 405 nm illumination. Such conclusion is mainly attributed to different resonant mechanisms of these two probes. The transverse and longitudinal oscillations of local electric field decide that the in-plane and out-of-plane field components, respectively, should be amplified by the PSL, which leads to the different optimization schemes. The proper PSLs can effectively amplify and compensate evanescent waves, resultantly improving the resolution, exposure depth and tolerance to the air gap of PDWL. Our work is expected to improve the shallow depth in practical PDWL and has the potential for sub-100 nm fabrication of photonic and plasmonic devices.